Discussion in 'Trading' started by areyoukidding?, Oct 5, 2005.
Anyone know of any?
Yes, how big? Many $5-10M or so blow ups, a few $100M+ blow-ups. Almost all can be attributed to poor risk management.
How did it happen? Position sizing?
No, one case I know of was incorrect hedging, so the system would keep on reset the pricing matrix (supposedly a start-up error by the Tokyo desk when they are done for the day), causing larger and larger "hedge" to be put into the market.
At time of discovery (when US risk picked up the unusual activity), the entity was accouning for 60% of all trades in the hedge instrument, pushing the price to unheard of levels. Suffice to say a few people got fired (including head of the japan desk), and the entity ended up down around $60-70M. That was a bad one.
Talk about a bad programming error. You would think someone given 100+MM would be able to put in risk parameters to counter that.
ahem, you will be surprised how primitive some of the systems are, especially since something that have been running for 10+ years. Problem isn't the original design, it is that the systems over the years got mangled to serve different purposes, a portfolio trading system that also take on index trading activities, etc, etc.
These systems are sufficiently large, with literally 20-30-40 components, not all of them are designed at the same time, often they are developed quite far apart.
Without mentioning names....something as silly (maybe even stupid) as placing a book on a keyboard sent enough orders out live to shut down Instanet for a period of time.....I would rate that as a "disaster" ....
Also, if the risk module was referring to the same pricing matrix, then the "hedge' sizes would appear to be "correct". Just that the risk module would not have the "common sense" to realize that such hedge sizes are "outragous" for that particular day. Since such trading can grow to very large sizes during portfolio rebalancing.
I don't know the exact specifics, I was put on the "fire-fighting squad" to clean up afterwards. But the worst part was that a member of the Tokyo desk wrote an e-mail "everything is a-ok" *after* the incorrect hedges started appearing. He was the first to go, obviously.
I think I remember the incident you are talking about, that was funny.
Yes however when you run a global operation you would think there would be independent risk management parameters that dont extract data from just one source alone.That's a disaster.